Reverted Phase Cancelation

[Ari-S]

I have know idea to explain this, so therefor this weird title...
But I was sampling a bass drum, and I had two instances of that bass drum. I cut two completely different parts of the track where the bassdrum was audible. Now when I align both samples and reverse the phase of one sample, it cancels out the bass drum of course, while the rest stay audible.
Now my question is: Is there some kind of technique or tool that actually does the opposite? So instead of deleting the stuff that is being phase cancelled, actually soloing it, so that I can extract only the stuff that is being cancelled out (in this case the clean bass drum).

Sorry if I explained this in a very bad way, but I tried my best
If someone actually understand what I'm talking about, can someone please tell me if it is possible? And if so, how?

[Burillo]

you basically have to feed the reversed bassdrum-less into the full signal

all - bass drum = bass drum less all
all - (all - bass drum) = bass drum

basic math.

[Ari-S]

you basically have to feed the reversed bassdrum-less into the full signal

all - bass drum = bass drum less all
all - (all - bass drum) = bass drum

basic math.

That doesn't work, because when I was canceling the bass drum out, I reversed the polarity one of the samples. So the "bassdrum-less" sample that was created as a result, still has a some reversed elements from that first reversed sample. So when adding the bassdrum-less sample to the total and reverse it, it will only cancel out all the stuff of the second sample, not of the first sample that had a reversed polarity.
(this is really f*cking up my mind, so again sorry for the bad explanation)

[aciddose]

To solo it you need to do something called a wavelet transform, then basically you convolve the signal with the sample of the kick. The signal will then become the correlation between the signal and the kick sample.

https://en.wikipedia.org/wiki/Cross-correlation

This won't leave you with a solo of the kick, but it will give you enough information to in most cases place the kick sample back in a new "resynthesized" signal.

https://en.wikipedia.org/wiki/Autocorrelation

You have heard of the "protection" in certain software where they claim a signal is "hidden, completely undetectable" that they could use to identify whether a pirated version were used in a recording? This likely relies upon autocorrelation techniques to extract the "hidden" signal, which is in fact not hidden at all. If it can be recovered, it isn't really hidden Only hidden to the untrained of course.

As far as what burillo mentioned though, entirely true.

if
A - B = C
then
C + B = A
and
C - A = -B

Try thinking of it this way:
C is +A and -B

so, to get A, you need to cancel out -B
B - B = nil, so adding +B: C + B = A

to get B, you need to cancel out +A
A - A = nul, so subtracting +A: C - A = -B


They call that sucker "algebra".

[Ari-S]

As far as what burillo mentioned though, entirely true.

if
A - B = C
then
C + B = A
and
C - A = -B

Try thinking of it this way:
C is +A and -B

so, to get A, you need to cancel out -B
B - B = nil, so adding +B: C + B = A

to get B, you need to cancel out +A
A - A = nul, so subtracting +A: C - A = -B


They call that sucker "algebra".

I understand it. But this is exactly my problem! How is going back to A or B giving me the clean bass drum? I mean, I already have A and B, So using C to go back to A or B again, won't make any sense in this case... The problem is that I actually want to extract the stuff that is being canceled out, but I have know idea how to do that. (btw, that wavelet transform method seems a little too complicated for me..)

[aciddose]

Yes of course, that was my point. It isn't possible for you to solo the bass drum.

(Assuming you don't already have the solo bass drum.)

If you already have A and B, then you can get C where B is cancelled out of A, but you already had A and B so extracting those from C doesn't make sense.

[camsr]

Yes it's possible, just phase invert the kick, mix with the original, then this new mix without the bass drum must be phase inverted, and mixed with another copy of the original.

[camsr]

you basically have to feed the reversed bassdrum-less into the full signal

all - bass drum = bass drum less all
all - (all - bass drum) = bass drum

basic math.

That doesn't work, because when I was canceling the bass drum out, I reversed the polarity one of the samples. So the "bassdrum-less" sample that was created as a result, still has a some reversed elements from that first reversed sample. So when adding the bassdrum-less sample to the total and reverse it, it will only cancel out all the stuff of the second sample, not of the first sample that had a reversed polarity.
(this is really f*cking up my mind, so again sorry for the bad explanation)

You did it wrong then. It works all day every day.

[Ari-S]

You did it wrong then. It works all day every day.

Let me explain it to you with audio files.
http://www56.zippyshare.com/v/RrCYzyQq/file.html
http://www56.zippyshare.com/v/GxebqTti/file.html

These are the two samples that I'm talking about. They're bot cut out of different parts of the track, so the only thing they have in common is the bass drum itself. So when I align them and invert the phase of one of them, only the bass drum will be canceled out.
So you get what I mean? I never had the clean bassdrum itself, so the method as explained earlier won't work. Like I said many times, I actually want the stuff that is being canceled out. But apparently there is no way to do this.

[Burillo]

if reversing one adding the two gives you the original without the kick drum, how about NOT reversing one of them? or, as i said, reversing the result and adding it back to the original signal.

[Ari-S]

if reversing one adding the two gives you the original without the kick drum, how about NOT reversing one of them? or, as i said, reversing the result and adding it back to the original signal.

As I said many times before, it doesn't work that way in this case. Grab the audio files and try it yourself.

[Ari-S]

double post....

[Burillo]

ah OK i get what you say now. i'm afraid you can't extract it if only audio had an AND operation...